Date of Award


Document Type

Campus Access Dissertation


Chemical Engineering

First Advisor

James A Ritter


The production and subsequent release of carbon dioxide into the atmosphere, no matter the source, is becoming an increasingly serious issue with respect to its affect on global warming. A considerable effort is underway worldwide to curb CO2 emissions from coal fired and other fossil fuel based power plants. Among many viable CO2 capture options available, Pressure Swing Adsorption (PSA) is gaining widespread industrial acceptance because of its attractive economics. The goal is to capture 90% of the CO2 from stack or flue gas, concentrate it to over 95 vol%, and sequester it somewhere in the Earth. The goal of this dissertation is to discuss and present the results obtained while developing a viable PSA process for CO2 capture, based on experiments and process modeling.

Most commercial PSA processes employ multiple beds that operate simultaneously. Each bed undergoes a series of cycle steps in order. During a cycle step, either a bed operates alone or it interacts with other beds. Such multi-bed, multi-step couplings give rise to different cycle configurations. The first four chapters of this dissertation introduce novel graphical and arithmetic approaches for complex PSA cycle scheduling.

Chapter 1 discusses a simple graphical approach aimed at deriving complex PSA cycle schedules. The methodology involves a priori specifying the cycle steps, their sequence, and the number of beds, and then following a systematic procedure that requires filling in a 2-D grid based on a few simple rules, some heuristics and some experience. The outcome or solution is a grid comprised of columns that represent the total cycle time, rows that represent the total number of beds, and cells that represent the duration of each cycle step, i.e., the complete cycle schedule. The methodology is successfully tested against several cycle schedules taken from literature, including a two-bed four-step Skarstrom cycle, a four-bed nine-step process with two equalization steps, a nine-bed eleven-step process with three pressure equalization steps, and a six-bed thirteen-step process with four pressure equalization steps and four idle steps. This approach reveals the existence of numerous cycle schedules for each bed and cycle step combination examined. Although it cannot identify the total number of permutations or which one is better, it does provide a very straightforward way to determine some of the possible cycle schedules of virtually any PSA process that can be conceived.

Chapter 2 discusses a novel unit block approach for rapid complex PSA cycle scheduling. The fundamental basis of this approach is very similar to the graphical approach discussed in Chapter 1. However, this methodology aims to teach a rather convenient and faster way of deriving PSA cycles, as the focus here is on a much smaller element of the cycle schedule called a unit block. This method involves a priori specifying the cycle steps, their sequence, and the number of beds, and then following a systematic procedure that requires filling in a 2-D grid. The outcome or solution is a unit block, which when extended forms the complete cycle schedule. This unique methodology can be applied to any number of beds, any number of cycle steps, and any number of coupled steps or constraints.

Multi-train PSA systems are typically operated and utilized for multi-component gas separations. Such PSA processes run at least two trains of PSA systems where each train consists of multiple beds and operates numerous cycle steps. The various cycle steps operating for a particular train can be arranged and configured in a variety of ways. In addition, there can be multi-bed interactions between different trains. Chapter 3 discusses a graphical approach for scheduling PSA cycles when multiple trains are involved based on the concepts introduced in Chapter 2. This new approach was tested successfully against several multi-bed and multi-step cycles taken from the literature that involve coupling of parallel trains. It should thus be very useful for quickly scrutinizing different PSA cycle schedules for further PSA process development.

The graphical methodologies discussed in Chapters 1, 2 and 3 do not give the total permutations or ways in which cycle schedules can be configured for a given number of beds, and operating sequence of cycle steps. Chapter 4 introduces an algebraic model for obtaining complex PSA cycle schedules. This new approach involves a priori specifying the cycle steps, their sequence and any constraints, and then solving a set of analytical equations. The solution identifies all the cycle schedules for a given number of beds, the minimum number of beds required to operate the specified cycle step sequence, the minimum number and location of idle steps to ensure alignment of coupled cycle steps, and a simple screening technique to aid in identifying the best performing cycles that deserve further examination. The methodology is tested successfully against 10, 12 and 16 bed PSA systems in the literature that all utilized the same 13 step cycle sequence that has four pressure equalization steps. It completely resolved all the corresponding cycle schedules for these 13 step multi-bed PSA systems with ease, and showed that the number of cycle schedules was hundreds to thousands of times greater than the few ever reported in the literature for each one.

Chapter 5 discusses an experimental study aimed at concentrating and separating CO2 from flue gas. A single bed PSA experimental system is utilized to measure thermodynamic and kinetic properties of CO2 and N2 on a commercial adsorbent. Adsorption isotherms and mass transfer coefficients extracted from breakthrough runs and pure gas cycling experiments are obtained. This information is used to validate an in-house developed adsorption process simulator named DAPS (Dynamic Adsorption Process Simulator). The single bed PSA experimental system is also utilized to study different PSA cycle schedules, because it mimics all the cycle steps of a complex multi-bed multi-step PSA process. Numerous runs are carried out over a range of conditions. They were used to further validate DAPS.

With a validated DAPS, Chapter 6 discusses a detailed parametric study carried out with the aim being to design an efficient PSA multi-bed multi-step process for capturing CO2 form coal fired power plant flue gas. The interplay and effects of various parameters like the purge to feed ratio, total cycle time, cycle schedules, and high to low pressure ratio on the overall process performance is studied. A number of PSA cycles are developed that achieved greater than 90% CO2 purity and 90% CO2 recovery, at high feed throughputs of greater than 2500 L STP/hr/kg. A direct advantage of such large feed throughputs is smaller bed sizes and less adsorbent inventory for new units and increased capability of handling large feed flow rates for existing units. For achieving 90% CO2 purity and recovery in the heavy product, energy consumption of 32.0 kJ/mol CO2 captured is calculated for a feed throughput of 2500 L STP/hr/kg. For a feed throughput of 2300 L STP/hr/kg, this number is 23.2 kJ/mol CO2 captured.

Overall this study showed that the Department of Energy's target of 90% CO2 purity and 90% CO2 recovery can be met by meticulously designing the PSA cycle and appropriate choice of operating parameters.